Modeling of Hydrogen Consumption and Process Optimization for Hydrotreating of Light Gas Oils

Transcription

1 Modeling of Hydrogen Consumption and Process Optimization for Hydrotreating of Light Gas Oils A Thesis Submitted to the College of Graduate and Postdoctoral Studies in Partial Fulfilment of the Requirements for the Degree of Master of Science in the Department of Chemical and Biological Engineering University of Saskatchewan Saskatoon By Adrian A. Rodriguez Pinos Copyright Adrian A. Rodriguez Pinos, May All rights reserved

2 PERMISSION TO USE It is my consent that the libraries of the University of Saskatchewan may make this thesis freely available for inspection. Besides, I agree that permission for copying of this thesis in any manner, either in whole or in part, for scholarly purposes be granted primarily by the professor(s) who supervised this thesis work or in their absence by the Head of the Department of Chemical Engineering or the Dean of the College of Graduate Studies. Duplication or publication or any use of this thesis, in part or in whole, for financial gain without prior written approval by the University of Saskatchewan is prohibited. It is also understood that due recognition shall be given the author of this thesis and to the University of Saskatchewan in any use of the material therein. Request for permission to copy or to make any other use of the material in this thesis in whole or in part should be addressed to: Head of the Department of Chemical and Biological Engineering 57 Campus Drive University of Saskatchewan Saskatoon, Saskatchewan, Canada S7N 5A9 i

3 ABSTRACT The main objective of this work was to develop a regression model for hydrogen consumption during hydrotreating of several gas oils such as virgin light gas oil (VLGO), hydrocracker light gas oil (HLGO), coker light gas oil (KLGO), and a partially hydrotreated heavy gas oil (PHTHGO) stream over commercial NiMo/ɣ-Al2O3 in a micro-trickle bed reactor. The experiments covered a temperature range of C, pressure range of MPa, LHSV range of h -1, and H2/oil ratio = 600 N m 3 /m 3. H2 consumption can be determined by different approaches; therefore, the best approach was selected by comparing the following: analysis of H2 content in gas streams, analysis of H2 content in liquid streams, and an approach reported in literature that is based on the decrease of the aromatics content. The comparison showed better agreement between the analysis of H2 content in gas streams and the method reported in the literature. For this reason, the analysis of gas streams was selected to build a regression model by performing statistical analysis of the effects of process conditions on H2 consumption data at the conditions mentioned above. H2 consumption based on gas analysis decreased in the following order: KLGO>VLGO>HLGO>PHTHGO. The H2 consumption regression model developed in this work was then tested with a new batch of experimental data, and the model performed better than similar correlations available in the literature. The secondary objective of this work was to study the effects of process conditions indicated above on hydrodesulfurization (HDS), hydrodenitrogenation (HDN), and hydrodearomatization (HDA) conversions by statistical analysis. The experimental data of hydrotreating conversions were then used to build regression models and carry out the optimization of process conditions for each feedstock. The optimum sets of conditions for the hydrotreating of each feedstock are the following: VLGO (T = 353 C, P = 9.37 MPa, LHSV = 0.9 h -1 ), HLGO (T = 383 C, P = MPa, LHSV = 0.9 h -1 ), KLGO (T = 372 C, P = 7.79 MPa, LHSV = 0.7 h -1 ), PHTHGO (T = 379 C, P = 9.44 MPa, LHSV = 0.7 h -1 ). In summary, the optimum hydrogen consumption as well as hydroprocessing conditions for all the four different feedstocks are substantially different. This information is critical in operating a commercial hydrotreater efficiently. ii

4 ACKNOWLEDGEMENT I would like to thank my supervisors Dr. A.K. Dalai and Dr. John Adjaye for their valuable input and guidance to take the project forward. Also, I appreciate the time and help provided by my advisory committee members: Dr. Catherine Niu and Dr. Hui Wang. I am greatly thankful to all the members of the Catalysis and Chemical Reaction Engineering Laboratory (CCREL) at the University of Saskatchewan for always being open to cooperate and assist with any inquiry related to this research project. Last but not least, I would like to appreciate the support provided by the technical and laboratory staff of the Department of Chemical and Biological Engineering. iii

5 DEDICATION This thesis is dedicated to my parents, Rosa Elena and Fernando, and my brother Luis for their support and motivation throughout the realization of this project. iv

6 TABLE OF CONTENTS PERMISSION TO USE... i ABSTRACT... ii ACKNOWLEDGEMENT... iii DEDICATION... iv TABLE OF CONTENTS..v LIST OF TABLES... ix LIST OF FIGURES... xi NOMENCLATURE AND ABBREVIATIONS... xiv 1 INTRODUCTION Background of the project Knowledge gaps Hypotheses Objectives Phase I- Development of hydrogen consumption regression models for hydrotreating of gas oils Phase II- To study the effects of process conditions on HDS, HDN, and HDA conversions and perform the optimization of process conditions during hydrotreating of gas oils Organization of the thesis LITERATURE REVIEW Bitumen recovery Bitumen upgrading Atmospheric distillation Vacuum distillation... 7 v

7 2.2.3 Hydrocracking Coking Hydrotreating Hydrotreating process variables Liquid hourly space velocity H2 partial pressure Temperature H2/oil ratio Hydrotreating chemistry Hydrodesulphurization Hydrodenitrogenation Hydrodearomatization Saturation of olefins Hydrotreating catalysts Hydrogen consumption Analysis of H2 concentrations in gas and liquid streams H2 consumption data reported in literature Stoichiometric H2 consumption in hydrotreating reactions Kinetic modelling Analysis of hydrogen consumption studies Vapor-liquid equilibrium in hydrotreating process Inlet and outlet hydrogen partial pressure Calculation of hydrogen partial pressure Dissolved hydrogen vi

8 3 EXPERIMENTAL Materials Experimental setup Experimental procedure Experimental plan Phase I- Development of H2 consumption regression models during hydrotreating of gas oils Phase II-Effects of process conditions on HDS, HDN, and HDA conversions during hydrotreating of gas oils and optimization of process conditions Characterization of feedstocks and products Boiling point distribution Nitrogen and sulfur analysis Analysis of aromatics content Measurements of hydrogen content Analysis of light hydrocarbons RESULTS AND DISCUSSION Development of hydrogen consumption regression models during hydrotreating of gas oils Hydrogen global mass balance Experimental hydrogen consumption Hydrogen consumption regression models Effects of temperature, pressure, and LSHV on hydrotreating conversions and optimization of process conditions during hydrotreating of gas oils Effects of temperature, pressure, and LSHV on HDA conversions vii

9 4.2.2 Effects of temperature, pressure, and LSHV on HDS conversions Effects of temperature, pressure, and LSHV on HDN conversions Summary of the effects of process conditions on hydrotreating conversions for gas oils Optimization of process conditions during hydrotreating of gas oils Effects of processing conditions on dissolved hydrogen, H2 inlet partial pressure, H2 outlet partial pressure, and feed vaporization during hydrotreating of gas oils Effects of outlet H2pp on feed vaporization, dissolved hydrogen and H2 consumption during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO SUMMARY, CONCLUSIONS AND RECOMMENDATIONS Summary Conclusions Recommendations..82 REFERENCES APPENDICES Appendix A: Experimental calibrations A.1 Hydrogen mass flow meter calibration A.2. Temperature calibration Appendix B: Additional experimental data viii

10 LIST OF TABLES Table 2.1 Typical composition of Athabasca bitumen... 6 Table 2.2 Atmospheric distillation cuts... 7 Table 2.3 Vacuum distillation cuts... 7 Table 2.4 Summary of operating conditions on hydrotreating of oil fractions Table 2.5 H2 consumption reported on literature Table 2.6 Laws explaining liquid-gas interaction for ideal mixtures Table 3.1 Properties of feedstocks Table 3.2 Experimental matrix Table 4.1 H2 consumption by gas analysis during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO Table 4.2 p-value test; H2 consumption during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO Table 4.3 H2 consumption regression models for VLGO, HLGO, KLGO, and PHTHGO Table 4.4 HDS, HDN, and HDA conversions, and H2 consumption during hydrotreating of LGOs mixture Table 4.5 Comparison between H2 consumption models for hydrotreating of petroleum fractions Table 4.6 HDA conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO Table 4.7 p-value test for HDA conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO Table 4.8 HDA regression models for VLGO, HLGO, KLGO, and PHTHGO Table 4.9 HDS conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO Table 4.10 p-value test for HDS activities during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO Table 4.11 HDS regression models for VLGO, HLGO, KLGO, and PHTHGO Table 4.12 HDN conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO.. 61 Table 4.13 p-value test for HDN conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO ix

11 Table 4.14 HDN regression models for VLGO, HLGO, KLGO, and PHTHGO Table 4.15 Overall effects of process conditions and interactions on hydrotreating conversions provided by the models Table 4.16 Optimum conditions for HDS conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO based on experimental data Table 4.17 Optimum conditions for HDN conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO based on experimental data Table 4.18 Optimum conditions for HDA conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO based on experimental data Table 4.19 Optimum process conditions for HDA alone during hydrotreating VLGO, HLGO, KLGO, and PHTHGO based on the regression models Table 4.20 Optimum process conditions to carry out hydrotreating of VLGO, HLGO, KLGO, and PHTHGO based on the regression models Table B.1 Inlet and outlet hydrogen partial pressure, feed vaporization, and dissolved hydrogen during hydrotreating of VLGO Table B.2 Inlet and outlet hydrogen partial pressure, feed vaporization, and dissolved hydrogen during hydrotreating of HLGO Table B.3 Inlet and outlet hydrogen partial pressure, feed vaporization, and dissolved hydrogen during hydrotreating of KLGO Table B.4 Inlet and outlet hydrogen partial pressure, feed vaporization, and dissolved hydrogen during hydrotreating of PHTHGO Table B. 5 Mass balance calculations and data for VLGO Table B. 6 Mass balance calculations and data for HLGO x

12 LIST OF FIGURES Figure 2.1 H2 cycle during hydrotreating Figure 2.2 Examples of sulfur compounds found in gas oils Figure 2.3 Schematic illustration of the hydrodesulphurization of thiophene Figure 2.4 Examples of basic and non-basic nitrogen compounds found in gas oils Figure 2.5 Schematic illustration of the hydrodenitrogenation of pyridine Figure 2.6 Schematic illustration of the hydrodearomatization of naphthalene Figure 3.1 Experimental setup Figure 4.1 Comparison of methods to calculate H2 consumption for HLGO Figure 4.2 Comparison of methods to calculate H2 consumption for KLGO Figure 4.3 H2 consumption comparison among feedstocks Figure 4.4 Effects of temperature on HDA conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Pressure and LHSV constant at 8.96 MPa and 1.5 h -1, respectively Figure 4.5 Effects of pressure on HDA conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Temperature and LHSV constant at 370 C and 1.5 h -1, respectively Figure 4.6 Effects of LHSV on HDA conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Temperature and pressure constant at 370 C and 8.96 MPa, respectively Figure 4.7 Effects of temperature on sulfur removal during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Pressure and LHSV constant at 8.96 MPa and 1.5 h -1, respectively Figure 4.8 Effects of pressure on sulfur removal during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Temperature and LHSV constant at 370 C and 1.5 h -1, respectively Figure 4.9 Surface plot (a) and Interaction plot (b) of effects of temperature and pressure on HDS conversions at LHSV= 1.5 h -1 for KLGO xi

13 Figure 4.10 Effects of LHSV on sulfur removal during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Temperature and pressure constant at 370 C and 8.96 MPa, respectively Figure 4.11 Interaction plot of the effects of temperature and LHSV on HDS conversions during hydrotreating of VLGO. Pressure constant at 8.96 MPa Figure 4.12 Effects of temperature on HDN conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Pressure and LHSV constant at 8.96 MPa and 1.5 h -1, respectively Figure 4.13 Effects of LHSV on HDN conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Temperature and pressure constant at 370 C and 8.96 MPa, respectively Figure 4.14 Effects of pressure on HDN conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Temperature and LHSV constant at 370 C and 1.5 h -1, respectively Figure 4.15 Interaction plots of (a) effects of temperature and LHSV on HDN conversions for KLGO at constant P (8.96 MPa), and (b) effects of pressure and LHSV on HDN conversions for KLGO at constant T (370 C) Figure 4.16 Interaction plots of the effects of pressure, temperature, and LHSV on HDN conversions for VLGO Figure 4.17 Interaction plot of the effects of temperature and LHSV on HDN conversions at constant P (8.96 MPa) for PHTHGO Figure 4.18 Effects of temperature on outlet and inlet H2pp for VLGO, HLGO, KLGO, and PHTHGO. Pressure and LHSV constant at 8.96 MPa and 1.5 h -1, respectively Figure 4.19 Effects of temperature on feed vaporization for VLGO, HLGO, KLGO, and PHTHGO. Pressure and LHSV constant at 8.96 MPa and 1.5 h -1, respectively Figure 4.20 Effects of pressure on feed vaporization for VLGO, HLGO, KLGO, and PHTHGO. Temperature and LHSV constant at 370 C and 1.5 h -1, respectively xii

14 Figure 4.21 Effects of temperature on dissolved hydrogen for VLGO, HLGO, KLGO, and PHTHGO. Temperature and LHSV constant at 370 C and 1.5 h -1, respectively Figure 4.22 Effects of pressure on dissolved hydrogen for VLGO, HLGO, KLGO, and PHTHGO. Pressure and LHSV constant at 8.96 MPa and 1.5 h -1, respectively Figure 4.23 Effects of hydrogen outlet partial pressure on dissolved hydrogen for VLGO, HLGO, KLGO, and PHTHGO Figure 4.24 Effects of hydrogen outlet partial pressure on hydrogen consumption for VLGO, HLGO, KLGO, and PHTHGO Figure 4.25 Effects of hydrogen outlet partial pressure on feed vaporization for VLGO, HLGO, KLGO, and PHTHGO Figure A. 1 Calibration curve for H2 mass flow meter Figure A. 2 Axial temperature profile Figure A. 3 Temperature calibration curve xiii

15 NOMENCLATURE AND ABBREVIATIONS Nomenclature Adj R 2 Modification of R 2 based on significant factors only API American Petroleum Insitute gravity CAr Aromatics content (wt %) cp Centipoise Cx Concentration of x species (ppm) G1 G2 G/O H/C Inlet gas stream flowrate (m 3 /h) Outlet gas stream flowrate (m 3 /h) Gas to oil ratio (Nm 3 H2/m 3 oil) Hydrogen to carbon ratio H 2G1 Concentration of hydrogen in gas inlet stream (mol %) H 2G2 Concentration of hydrogen in gas outlet stream (mol %) H 2L1 Concentration of hydrogen in liquid feedstocks (mol %) H 2L2 Concentration of hydrogen in liquid products (mol %) H2pp Hydrogen partial pressure (MPa) L1 L2 Inlet liquid stream flowrate (m 3 /h) Outlet liquid stream flowrate (m 3 /h) LHSV Liquid hourly space velocity (h -1 ) N Normal laboratory conditions (P= kpa and T=20 C) P Pressure (MPa) R 2 T V Statistical measure that represents the goodness fit of a model Temperature ( C) Volumetric flow (m 3 /h) xiv

16 Abbreviations CCD HDA HDN HDS HLGO KLGO LGOs PHTHGO VLE VLGO Central composite design Hydrodearomatization Hydrodenitrogenation Hydrodesulfurization Hydrocracker light gas oil Coker light gas oil Light gas oils Partially hydrotreated heavy gas oil Vapor-liquid equilibrium Virgin light gas oil xv

17 1 INTRODUCTION The use of bitumen-derived gas oils is indispensable because of the increasing demand for oil and the limited availability of conventional oil sources. Conventional oil sources are expected to decrease to 60 % in 2040 as compared to 80 % in 2010 (World Energy Outlook, 2011). Hydrotreating is one of the essential processes to upgrade bitumen derived gas oils by lowering sulfur, nitrogen and aromatic contents. An oil market forecast performed in the year 2010 predicted that by 2013, hydrotreating units would produce million bbl oil/day out of 90.8 million bbl oil/day produced worldwide (Silvy, 2010). The production capacity of this process implies its significant contribution to the global hydrogen mass balance of refineries During primary upgrading of bitumen, light gas oils (LGOs) can be obtained from various processing units. Virgin light gas oil (VLGO) is obtained from the atmospheric distillation of bitumen dissolved in naphtha, vacuum distillation of the atmospheric distillation unit residue produces vacuum LGO (LVGO). Coker LGO (KLGO) and hydrocracker LGO (HLGO) are obtained from the conversion of vacuum residue in coking and hydrocracking units, respectively (Jones & Pujado, 2006). Because of the varying composition and properties of hydrotreating feedstocks, special attention has to be paid to study the variability in H2 consumption during the processing of each feed for technical and economical purposes. 1.1 Background of the project Hydrogen is one of the most important elements in refinery operations due to its high demand and price. Gas oil hydrotreaters use hydrogen in excess to prevent the formation of coke and provide an optimal liquid-solid contact (Ancheyta & Speight, 2007).These units consume about N m 3 H2/m 3 oil (Ramachandran & Menon, 1998). The outlet gas stream from hydrotreaters is composed of H2S, NH3, C1-C6, and H2 with a concentration of mol. % (Peramanu et al., 1999). In the industry, this stream is purified to recover the excess of H2 and 1

18 then it is recycled and mixed with pure H2 to achieve the required hydrogen partial pressure (H2pp) in the hydrotreater and make up for the hydrogen consumed in the different hydrotreating reactions. Hydrogen consumption is dependent on feedstock properties. Feedstocks with lower hydrogen-carbon ratio (H/C) such as KLGO (H/C = 0.13) in comparison with VLGO (H/C = 0.15), will demand more hydrogen to improve the degree of unsaturation (Yui, 2008). Similarly, higher amounts of sulfur and nitrogen impurities will also translate into more H2 demand. The total H2 consumption in the industry not only involves the amount of H2 consumed in hydrotreating reactions, but also considers mechanical losses due to the use of compressors, venting losses during gas purging, and solution losses (dissolved H2) in the final product (Hisamitsu et al., 1976). In the case of experimental studies, dissolved H2 and chemical H2 consumption should be considered; dissolved H2 may affect total H2 consumption by a factor of 6 % (Castañeda et al., 2011). Dissolved H2 in hydrotreated products can be calculated in HYSYS software by performing VLE (vapor-liquid equilibrium) simulations (Mapiour et al., 2010). Two important issues may arise to determine H2 consumption in hydrotreating processes. First, developing hydrogen mass balances and the use of methods to determine chemical H2 consumption based on reaction stoichiometry require the use of different analytical techniques such as gas chromatography (GC), GC coupled to mass spectroscopy, elemental analysis (measures C, H, N, and S content) and 13 C nuclear magnetic resonance (Castañeda et al., 2011). Second, values and models available in literature have limited application. For instance, Castañeda et al. (2011) found out that Edgar s approach (Edgar, 1993) presented an error higher than 50 % for the determination of H2 consumption during hydrotreating of naphtha and gas oils. Jones & Pujado (2006), reported values of and N m 3 H2/m 3 oil for each wt. % change in sulfur or nitrogen content, respectively. Meanwhile, Gary et al. (2007), reported values of 12.5 and 57 N m 3 H2/m 3 oil for each wt. % change in sulfur or nitrogen content, respectively. 1.2 Knowledge gaps The amount of hydrogen consumption during hydrotreating is important for refineries to determine and calculate the overall hydrogen mass balance. Problems may arise when these hydroprocessing units process feedstocks with varying composition and properties, which may 2

19 require different amounts of hydrogen. The overestimation of hydrogen may cause economic losses, and underestimating the amount of this reactant may cause operational problems and possible shutdown of the unit. It is well known that modern refineries treat numerous feedstocks including fractions from residues to maximize the output from crude petroleum feedstock. Therefore, the knowledge to determine hydrogen consumption for different feedstocks in essential. There are several models in the literature to determine H2 consumption during hydrotreating; however, these models may present disadvantages such as the following: the models are feed dependent and therefore cannot be used for all petroleum fractions, models developed based on reaction stoichiometry require detailed characterization of feedstocks and products, and these models omit dissolved H2. In addition, the values of H2 consumption from these models are too general for specific application. Based on these arguments, there is the need to carry a comparative study for hydrogen consumption between common hydrotreating feedstocks considering two main purposes: (i) provide scientific explanation on the behavior of each feedstock towards hydrogen consumption at industrial operating conditions during hydrotreating, and (ii) develop simple models to calculate H2 consumption based on processing conditions. 1.3 Hypotheses - Hydrogen consumption is dependent on process parameters; thus, statistical models based on process conditions can provide an accurate and simple prediction of this dependent variable. - Each feedstock has a different set of optimum conditions of temperature, pressure, and LHSV to maximize HDS (hydrodesulfurization), HDN (hydrodenitrogenation), and HDA (hydrodearomatization) conversions, which can provide data for the development of statistical models to predict HDS, HDN and HDA for these gas oils. 1.4 Objectives The main objective of this work was to develop a H2 consumption regression model applicable to the hydrotreating of light gas oils such as KLGO from coking, HLGO from hydrocracking, VLGO from atmospheric distillation, and PHTHGO from a hydrotreating unit 3

20 based on temperature, pressure, and LHSV. The developed model was compared with other models and data available in the literature. The second objective of this work was to study the effects of process conditions on hydrotreating conversions for the light gas oils mentioned above and find the best sets of conditions to process each feedstock. Based on these two objectives, this work was divided into two phases Phase I - Development of hydrogen consumption regression models for hydrotreating of gas oils Hydrogen consumption was measured by liquid and gas stream analysis and the best method was selected to build regression models based on process conditions. The experiments were designed by central composite design (CCD). Based on the data for each experimental run and feedstock, a composite regression model of H2 consumption was built and then the model was tested with a new batch of experimental data. The response of model was compared with other models and data available in the literature.the models available in the literature calculate chemical H2 consumption only; therefore, dissolved H2 was added to the chemical H2 consumption. Dissolved H2 was calculated in HYSYS software by performing VLE (vapor-liquid equilibria) calculations. Data of H2pp and feed vaporization can also be obtained from the simulation; therefore, this work also studied the effects of process conditions on H2 inlet pp, H2 outlet pp, and feed vaporization Phase II Development of the statistical models for HDS, HDN, and HDA conversions based on the process parameters and perform the optimization of process conditions during hydrotreating of gas oils Hydrotreating conversions were determined for the experiments given by the CCD to examine the single and combined effects of process conditions on HDS, HDN, and HDA. Consequently, the conversion data were analyzed statistically to develop regression models and optimize process conditions for each feedstock. 4

21 1.5 Organization of the thesis In addition to the Introduction chapter, this thesis contains the following chapters. Chapter 2 explains the concepts and principles of the processing route to convert bitumen into hydrotreated oil and the challenges to determine H2 consumption during hydrotreating. Chapter 3 describes the materials and experimental methods which were used to carry out each of the phases proposed in this thesis. Chapter 4 presents a comprehensive and detailed discussion of the results obtained in this research work. Chapter 5 provides a summary of the main findings of this research as well as conclusions and recommendations for the future work. 5

22 2 LITERATURE REVIEW This chapter describes the principles and concepts for extracting hydrotreated oil from bitumen. Major emphasis was placed on describing the methods to determine H2 consumption in hydrotreating and the challenges associated with the usage of the various models and data available in the literature. Also, this chapter describes the process optimization studies of various gas oils for HDS, HDN, and HDA. 2.1 Bitumen recovery The components of oils sands are bitumen (organic components), quartz/clay particles, and water. Oil sands are produced by surface mining when the reservoirs are located at maximum of 75 m of depth. Then, bitumen is recovered by treating oil sands with alkaline hot water. When the reservoirs of oil sands are situated below 75 m, bitumen is recovered by in-situ methods such as steam assisted gravity drainage (SAGD) and cyclic steam simulation (CCS) (Banerkke, 2012). Bitumen has a viscosity greater than 10,000 cp and API gravity less than 10 in comparison with other unconventional oils (1,000-10,000 cp and < 10 API) and conventional oil sources (1-1,000 cp and API) (Banerkke, 2012). The typical composition of bitumen in the Canadian Athabasca deposits is presented in Table 2.1. Table 2.1 Typical composition of Athabasca bitumen (Banerkke, 2012) Carbon wt. % Hydrogen wt. % Nitrogen 3,000-5,000 ppm Sulfur wt. % Oxygen < 1.0 wt. % Vanadium Nickel ppm ppm 6

23 2.2 Bitumen upgrading Atmospheric distillation Bitumen dissolved in naphtha is distilled in a pressure range of 5-10 psig to obtain the products shown in Table 2.2 (Jones & Pujado, 2006). The residue can be processed either by vacuum distillation or thermal cracking. Table 2.2 Atmospheric distillation cuts (Jones & Pujado, 2006) Products Temperature range ( C) Naphtha Kerosene VLGO Virgin heavy gas oil (VHGO) Residue (Fuel oil) > Vacuum distillation The residue of the atmospheric distillation unit is processed by vacuum distillation at approximately 1.3 kpa to obtain light vacuum gas oil (LVGO), heavy vacuum gas oil (HVGO), and topped bitumen as shown in Table 2.3. Vacuum is used to prevent cracking of the feed (Jones & Pujado, 2006). Table 2.3 Vacuum distillation cuts (Jones & Pujado, 2006) Products Temperature range ( C) LVGO HVGO Topped bitumen >

24 2.2.3 Hydrocracking Hydrocracking is a high pressure (13.79 MPa) catalytic process used for cracking long chain hydrocarbons in the presence of hydrogen. It also performs HDA, HDS, HDN and HDM (hydrodemetallization) function to prevent the poisoning of the cracking catalyst. Hydrocracking can process several feeds such as vacuum gas oils, straight run gas oils, thermally cracked gas oils, and vacuum distillation bottoms. (Ancheyta & Speight, 2007). Usually, the products from hydrocracking do not require further upgrading. Hydrocrackers consume approximately N m 3 H2/m 3 oil (Jones & Pujado, 2006). LC-fining is a type of hydrocracking process that has HDS (60-90 %), and HDM (50-98 %) function. This process features an expanded catalyst bed which is in constant movement due to the up-flow of oil stream. Moreover, the catalyst is added and removed from the reactor after certain periods of time allowing the process to run continuously. The main advantages of this process are the following: low operating and investment costs, greater recovery of light fractions, and high yield of liquid products. The operating conditions of the process are the following: T = C, and P = MPa. Moreover, the LC-fining process consumes about N m 3 H2/m 3 oil (Rana et al., 2007) Coking Coking uses only heat to crack heavy oil fractions or residues and convert them into lighter products such as naphtha, coker light gas oil (KLGO), and coker heavy gas oil gas oils (KHGO). There are two major coking processes: delayed coking and fluid coking. In the delayed coking process, the feed is preheated in a heat exchanger. Then, the furnace heats the feed which is retained in a coil for a determined period of time (up to 24 h) to temperatures higher than its cracking temperature (493 C). The product that leaves the furnace is then directed to several coker drums in which the coking temperature is maintained. Then, the cracked oil is separated from the coke by cooling it down below the cracking temperature. Finally, the oil is sent to a fractionator (Jones & Pujado, 2006). In the case of fluid coking, the feed is cracked using the heat generated by a coke bed fluidized by steam. The advantages of fluid coking over delayed coking are a higher yield of products and less consumption of energy and utilities; however, the products are of low quality. An important type of fluid coking process in the industry is Flexicoking. The latter uses a gasifier 8

25 to convert some fraction of the coke into syngas; however, the coke cannot be converted completely into oil fractions even at temperatures as high as 1000 C (Jones & Pujado, 2006; Rana et al., 2007). Since coking processes operate at high temperatures, the paraffins and cyclic paraffins that are usually found in crude oil are transformed into olefins due to unsaturation (Rana et al., 2007). Moreover, the products from coking contain a high percentage of aromatic compounds (Aoyagi et al., 2003). 2.3 Hydrotreating Petroleum fractions obtained from unconventional oil sources contain high amounts of impurities such as sulfur, nitrogen, oxygen and in some cases metalloporphyrins. In addition, large amounts of aromatic compounds are present in gas oils. The reaction between oil and hydrogen takes place in a catalyst bed at high temperatures and pressures in a process called hydrotreating. Hydrotreating is capable of reducing sulfur, nitrogen, and aromatic contents to comply with environmental regulations and improve the quality of the final product (Papayannakos & Georgiou, 1988). Hydrotreating is dependent on the feed properties, the catalyst type and the operating conditions of the process (Ancheyta et al., 2005). Hydrotreating is performed in a continuous trickle bed reactor operating close to a plug-flow in which the reactants (gas oil and hydrogen) are introduced from the top of the reactor. The reactants come in contact with the catalyst bed and HDS, HDN and HDA reactions take place.the gas and liquid products pass through a water scrubber to remove ammonia before going to a highpressure separator. The liquid is collected and fractioned to obtain the final products; meanwhile, the gas stream proceeds to a series of units for purification (Gary et al., 2007). The non-condensable gases are composed of mol. % H2, C1-C5, NH3 and H2S; the light ends (C1-C5) are produced due to cracking of the feed (< 20 vol. % of the feedstock) (Jones & Pujado, 2006). This stream is then taken to a sour water plant unit (amine contactor) to separate H2S with the purpose of enhancing H2 purity and preventing the release of H2S to the environment. Following this step, % of the gas stream is purged to the fuel gas system or to a H2 purification process and the rest of the gas stream rich in H2 (80-85 mol. % H2) is compressed and mixed with make-up H2 ( mol. % H2) (Peramanu et al., 1999; Jones & Pujado, 2006). Figure 2.1 shows the cycle of H2 in the industry after hydrotreating. 9

26 Make-up H2 HYDROTREATER Off-Gas mol. % H2 H2S C1-C5 AMINE UNIT HYDROGEN RECOVERY UNIT Compressor Tail gas H2S dissolved H2 Chemical H2 Consumption AMINE UNIT Fuel gas Figure 2.1 H2 cycle during hydrotreating (Peramanu et al., 1999) H2S 2.4 Hydrotreating process variables The main independent process variables of hydrotreating are liquid hourly space velocity (LHSV), temperature, H2 partial pressure and H2/oil ratio. Table 2.4 provides a summary of the operating conditions used for hydrotreating of different petroleum fractions Liquid hourly space velocity This variable is defined as the ratio between the volumetric flow rate of the feedstock and the volume of the catalyst used for the operation. Lower LHSV implies an increase in the residence time of the gas oil in the catalyst bed and it generally improves hydrotreating activities. LHSV can be set to maintain optimal conversion levels at high temperatures. However; the temperature cannot be increased to the point at which coke is formed. This will result in catalyst deactivation and consequently, the catalyst life will be reduced (Ancheyta & Speight, 2007). 10

27 2.4.2 H2 partial pressure Hydrogen partial pressure is the product of process pressure and H2 purity. High H2pp enhances the performance of the hydrotreating process by increasing the catalyst life and the throughput capability. Also, it allows the process to handle heavier feeds, increase the quality of the distillate, and promotes the elimination of the purged gas (Jones & Pujado, 2006). More details about H2 partial pressure will be mentioned in the vapor-liquid equilibrium section Temperature High temperature increases the removal of impurities; however, a temperature higher than 410 C promotes thermal cracking of the feed and produces hydrocarbons of low molecular weight. Moreover, high temperatures also deactivate the catalyst by coke formation (Ancheyta & Speight, 2007) H2/oil ratio Hydrogen is used in excess (4 times more than the chemical H2 consumption) to maintain an adequate contact between the feed and the catalyst which promotes the conversion of the reactants, removal of impurities and at the same time reduces the possibility of coke formation (Ancheyta & Speight, 2007). To optimize the use of H2 which is costly, the use of recycled gas with addition of high purity H2 is essential to maintain the hydrogen partial pressure. Hydrotreating units consume around N m 3 H2/m 3 oil (Ramachandran & Menon, 1998). 11

28 12 Table 2.4 Summary of operating conditions on hydrotreating of oil fractions Study Catalysts and operating conditions Summary Kinetics of Commercial NiMo/Al2O3 At higher residence time, HDS activities increased. HDS conversion hydrodesulfurization of heavy T = C of 96 wt% was obtained at LHSV = 0.5 h -1. gas oil derived from oil-sands P = 8.8 MPa Temperature had a positive effect on HDS activities. This effect bitumen. LHSV = h -1 started to plateau at a temperature range of C. ( Bej et al., 2007) H2/oil = m 3 /m 3 HDS activities did not show any significant improvement at H2/oil Feedstock: Heavy gas oil ratio higher than 800 m 3 /m 3. Hydrogenation of coker Three NiMo catalysts The three catalysts did not show any significant difference in terms naphtha with NiMo catalyst. T = C of activities. (Yui & Chan, 1992) P = 3-5 MPa Temperatures higher than 220 C had a positive effect on HDS and LHSV 1.0 and 2.0 h -1 HDN activities, while HDA conversions improved at values higher H2/oil = 600 m 3 /m 3 than 200 C. Feedstock: Coker Naphtha Pressure did not show any significant effect on any of the hydrotreating conversions. Residence time had a significant effect for HDS, HDN, and HDA. Product Selectivity during Commercial NiMo/Al2O3 Temperature had a positive effect on HDS and HDN conversions. Hydrotreating and Mild T = C Sulfur contents reached lower values than nitrogen contents; Hydrocracking of Bitumen- P = MPa however, temperature was more significant on HDN conversions. Derived Gas Oil LHSV = h -1

29 13 (Botchwey et al., 2003) H2/oil = 600 m 3 /m 3 Feedstock: Heavy gas oil Pressure improved HDN activities more significantly than HDS activities. As LHSV increased, HDS and HDN conversions decreased. Higher values of LHSV improved the selectivity of sulfur removal over nitrogen removal. Temperature and pressure improved HDA conversions. Pressure was more significant on HDA conversions at lower temperatures due to equilibrium limitations. Catalytic hydrorefining of heavy gas oil. (Mann et al., 1987) Ni-Mo, Ni-W, and Co-Mo catalysts supported on alumina T = C LHSV = h -1 P = MPa HDN and HDS conversions increased linearly with temperature. Ni-Mo catalyst had a better performance at temperatures lower than 350 C; meanwhile, Ni-W was better at temperatures higher than 350 C. Pressure had a positive effect on HDS and HDN conversions. This process condition was more significant on HDN conversions. Higher residence time improved HDS and HDN conversions. Changes in residence time were more significant on HDN. C/H ratio was not affected by LHSV for experiments using NiMo catalyst at T = 450 C and LHSV of 2 h -1. This means HDA conversions were independent of changes in the residence time due to equilibrium limitations.

30 14 Hydrotreating of straight run gas oil light cycle oil blends (Ancheyta-Juárez et al., 1999) Two stage aromatics hydrogenation of bitumenderived light gas oil (Owusu-Boakye, 2005) Co Mo/γ-Al2O3 T = 350, 360,370, 380 C LHSV = 1, 1.5, 2 h -1 P = 5.29,6.86,8.82 MPa H2/oil = ml/ml Feeds: -Blend A: 80 vol. % straightrun gas oil (SRGO), and 20 vol. % light cycle oil (LCO). -Blend B: 50 vol.% SRGO, and 50 vol.% LCO. Single stage study summary: NiMo/Al2O3 T = C P = MPa LHSV = h -1 H2/oil = 550 ml/ml C/H ratio was not improved significantly with increasing temperature. On the other hand, pressure had a positive effect on C/H ratio. LCO has lower cetane index (CI) and higher contents of S, and N than SRGO. Blend B needed higher reaction temperature to achieve the desired S content (<500 wppm) due to higher initial S content. At higher temperatures S, N, and aromatics content decreased. Consequently, API and CI increased. Lower LHSV enhanced the product quality. Increasing pressure enhanced the product quality and the HDS and HDA rate. Maximum HDA conversion of 69 % at T = 379 C, P = 11.0 MPa, and LHSV= 0.6 h -1 HDA was affected primarily by temperature. HDS and HDN conversions were affected by temperature and also by pressure. Increasing temperature and pressure improved the CI. In the case of temperature, CI reached an equilibrium.

31 15 Feed: light gas oil blend composed of vacuum light gas oil (VLGO), atmospheric light gas oil (ALGO), and hydrocracker light gas oil (HLGO). HDS reactivities of NiMo/Al2O3 HDS activities increased with increasing temperature. dibenzothiophenic compounds Catalyst volume = 150 ml 4, 6- DMDBT and 4-MDBT had the lowest and highest HDS activity in an LC-finer LGO and H2S/NH3 inhibition effect (Chen & Ring, 2004) T = C P =5-11 MPa LHSV = h -1 Gas rate = NL/h of all the sulfur compounds studied, respectively. 1, 4, 6-TMDBT and 2, 4, 6-TMDT had higher activity than 4, 6- DMDBT showing that methyl groups at carbon positions 1 or 2 may increase HDS activities due to possible contributions of the hydrogenation mechanism.

32 2.5 Hydrotreating chemistry The reactions that are of greater concern during hydrotreating of oils are HDS, HDN, and HDA. The H2 consumption in each reaction is shown in Table 2.5. The ease of carrying each reaction follows the next order: HDS>HDN>HDA (Jones & Pujado, 2006). Olefins saturation may also be present, especially during hydrotreating of feedstocks produced in thermal processes. Table 2.5 H2 consumption reported on literature Reaction H2 consumption / wt. % change (N m 3 H2/m 3 oil) HDS HDN HDA 4.0(per vol. %) 4.81 Reference Jones & Pujado, 2006 Gary et al., 2007 Ancheyta & Speight, 2007 Edgar,1993 Jones & Pujado, 2006 Gary et al., 2007 Ancheyta & Speight, 2007 Edgar, 1993 Ancheyta & Speight, 2007 Edgar, Hydrodesulphurization Hydrodesulphurization (HDS) is an exothermic and generally irreversible reaction under industrial operating conditions ( C and MPa) (Girgis & Gates, 1991). There are six types of sulfur compounds and its derivatives that can be present in different distillate products. These are mercaptans, sulfides, disulfides, thiophenes, benzothiophenes, and dibenzothiophenes. In the case of gas oils, most of the sulfur compounds are present as benzothiophenes and dibenzothiophenes ( Jones & Pujado, 2006; Ancheyta & Speight, 2007). Some examples of sulfur compounds found in oil are shown in Figure

33 Figure 2.2 Examples of sulfur compounds found in gas oils HDS can proceed by two possible mechanisms: direct hydrogenolysis of the C-S bond or hydrogenation of an unsaturated bond followed by hydrogenolysis which will consume more hydrogen than the first mechanism. In contrast to hydrogenolysis, hydrogenation is dependent on hydrogen partial pressure (Girgis & Gates, 1991). Both mechanisms are illustrated in Figure 2.3. Furthermore, the reactivity of benzothiophenes decreases with increasing molecular weight e.g. BT (benzothiophenes) are less reactive than DBT (dibenzothiophenes) (Stanislaus et al., 2010). When methyl groups are connected to the benzenoid ring in positions 4 and 6, the reactivity also decreases. In contrast, methyl groups in positions 2, 8, 3, or 7 do not show any significant change in the reactivity (Girgis & Gates, 1991). Sulfur compounds in oil fractions are numerous e.g. light gas oil from an Arabian crude oil was found to have at least 42 alkyl-substituted benzothiophenes and 29 alkyl-substituted dibenzothiophenes (Ishihara et al., 2005). Figure 2.3 Schematic illustration of the hydrodesulphurization of thiophene Hydrodenitrogenation The removal of nitrogen is an exothermic reaction which is generally irreversible under industrial operating conditions (Jones & Pujado, 2006). The nitrogen compounds found in oil are 17

34 generally formed of five to six unsaturated rings and are classified as non-heterocyclic and heterocyclic compounds. Non-heterocyclic compounds (anilines, aliphatic amines, and nitriles) are de-nitrogenated more rapidly than heterocyclic compounds (basic: quinoline, acridine; nonbasic: indole, carbazole) However, non-heterocyclic compounds are present in oil in small quantities (Girgis & Gates, 1991; Ancheyta & Speight, 2007). The reactivity of heterocyclic nitrogen compounds seems to be independent of the number of benzenoid rings in the molecule. This means that the effect of the steric hindrance caused by the presence of various aromatic rings is independent of the molecular size for the C-N scission (Girgis & Gates, 1991). Furthermore, non-basic compounds are less reactive than basic compounds found in real oil feedstocks when temperature and pressure increases. This could be due to the prevailing effect of pre-hydrogenation reactions during the process which transforms non-basic into basic compounds (Bej et al., 2001). Figure 2.4 Examples of basic and non-basic nitrogen compounds found in gas oils HDN mechanism proceeds in three steps: hydrogenation, hydrogenolisis, and denitrogenation. Hydrogenation is required to facilitate the removal of nitrogen because the amount of energy to break the -N=C bond (147 kcal/mol) is higher than the energy required to break the C-N- bond (73 kcal/mol) (Girgis & Gates, 1991; Jones & Pujado, 2006). The removal of nitrogen from pyridine is shown in Figure

35 Figure 2.5 Schematic illustration of the hydrodenitrogenation of pyridine Hydrodearomatization Hydrodearomatization is an exothermic reaction and it can be reversible under normal industrial operating conditions as shown in Figure 2.6. Hydrogen partial pressure favors this reaction and high temperatures limit it (Girgis & Gates, 1991). According to Mapiour et al. (2010), the conversion of aromatics reaches its maximum activity at C. The compounds found in gas oils may contain a maximum of three aromatic rings. Most of the aromatic compounds are not completely saturated under normal hydrotreating conditions but converted to naphthenic rings. The saturation of aromatics improves the properties of oil products such as cetane index and smoke point (Jones & Pujado, 2006).The reactivity of aromatic compounds increases with the presence of more aromatic rings while the methyl substituents do not affect the reactivity to a great extent. The greater reactivity of polyaromatics might be due to greater resonance stability of the π bond between naphthalene fractions and the active sites of the catalysts (Girgis & Gates, 1991). A simple illustration of hydrodearomatization reaction is shown in Figure 2.6. Figure 2.6 Schematic illustration of the hydrodearomatization of naphthalene 19

36 2.5.4 Saturation of olefins Olefins are not found naturally in petroleum; however, they can be formed during thermal or catalytic processes such as coking, ethylene manufacturing, among others. Olefins saturation is an exothermic reaction (Jones & Pujado, 2006). 2.6 Hydrotreating catalysts Catalysts used for hydrotreating of heavy oils require an optimal balance between high activities and proper physical properties. High surface area and moderate pore volume improve HDS activities due to high dispersion of the active phases. However, feeds with high metal contents will require catalysts with higher pore volume to prevent mouth plugging (Ancheyta et al., 2005). Also, due to processing at high temperatures, coke can be formed and dispersed between the pores of the catalyst causing deactivation (Ancheyta & Speight, 2007). The most widely used catalyst in the industry is composed of molybdenum supported on gamma alumina (γ-al2o3) and promoted by either nickel or cobalt (Papayannakos & Georgiou, 1988). The promoters function is to enhance the selectivity and activity of the catalysts by decreasing the metal-sulfur interaction in MoS2. Ni prefers to be attached to the Mo edge providing higher hydrogenation activities, while Co prefers the S-edge providing higher sulfur removal activities (Sun et al., 2005). The active phases (Co-Mo-S or Ni-Mo-S) of the catalyst are produced during sulfidation. When interactions between active phases and the support are weak, the active phases form a stack of 2-4 layers defined as Type II Co-Mo-S or Ni-Mo-S (Topsøe, 2007). From this species, the sulfur ion is expelled as H2S forming sulfur vacancies commonly known as coordinatively unsaturated sites (CUS) with Lewis acid nature. This property allows the adsorption of molecules with unpaired electrons such as pyridine and DBT. On the neighboring site, the heterocyclic and homocyclic splitting of hydrogen form Mo-H and SH groups which are the sources of hydrogen for the hydrotreating reactions to occur. These groups have Bronsted acid character (Ancheyta et al., 2005). The reaction completes by the re-formation of the original active phases and it is known as sulfur-breathing. Mochida & Choi (2006 described the sulfur-breathing mechanism graphically. Alumina is the support of choice due to characteristics such as metal oxides from group VI, and VIII oxides are highly dispersed forming inactive compounds, stabilization of the operating 20

37 phases, high purity, physical properties can be tailored easily, thermal stability and regeneration conditions, malleable into different forms, and low cost (Luck, 1991). Supports can either be modified chemically or physically to improve their characteristics; additives such as phosphate have shown improvement on activity, and the addition of TiO2 to alumina allowed control of pore volume and mean pore diameter. Research has also been focused on testing other types of supports such as mixed oxides, and carbon-based supports for improving acidic sites distribution for a better hydrocracking activity, and improving the resistance to N bases, respectively (Ancheyta et al., 2005). Other physical properties of concern are mechanical strength, and sizes and shapes of catalysts particles. Sufficient mechanical strength is required to prevent breakage during the process, and specific sizes are required to prevent diffusion problems; large particles may present problem of underutilization while small particles (under 0.8 mm) may cause pressure drop in the reactor. The most common shapes are sphere, pellet, cylinder, bilobular, trilobular and teralobular respectively (Ancheyta et al., 2005; Rana et al., 2007). 2.7 Hydrogen consumption The catalyst properties, the level of conversion, operating conditions and properties of the feedstock affect H2 consumption during hydrotreating (Castañeda et al., 2011). Therefore, the determination of H2 consumption is important for economic and technical issues. Pilot plant studies consider chemical H2 consumption and dissolved H2 only; studies at pilot scale can generally obtain accurate values by performing twenty experiments (Lee et al., 2008). Hydrogen is present in the liquid and gas streams of the process. Thus, the overall mass balance for hydrogen takes the following form: G 1 H 2G1 + L 1 H 2L1 = G 2 H 2G2 + L 2 H 2L2 (2.1) 21

38 where: G1 and G2 are the gas inlet and outlet flow rates (m 3 /h), respectively; H 2G1 and H 2G2 are the hydrogen concentrations in gas inlet and outlet streams, respectively (mol. %); L1 and L2 are the liquid inlet and outlet flow rates, respectively (m 3 /h); H 2L1 and H 2L2 are the hydrogen concentration in feedstocks and products, respectively (mol. %). The hydrogen concentration in the liquid products takes into consideration unreacted dissolved H2 as well as molecular H2 added to the products due to reaction. In general, there are five approaches to determine H2 consumption during hydrotreating. These approaches are listed below. 1. Analysis of H2 concentrations in gas streams. 2. Analysis of H2 concentrations in liquid streams. 3. H2 consumption data reported in the literature. 4. Stoichiometric H2 consumption in each type of reaction. 5. Kinetic modeling Analysis of H2 concentrations in gas and liquid streams The analysis of H2 contents in gas and liquids require the use of different analytical techniques e.g. elemental analysis or nuclear magnetic resonance (NMR) for analysis of liquids, and high-resolution gas chromatography for analysis of gases (Castañeda et al., 2011). The equations used for both methods are shown below. H 2 consumption gas (N m3 H 2 m 3 ) = G 1 H 2G1 - G 2 H 2G2 oil L 1 (2.2) H 2 consumption liquid (N m3 H 2 m 3 ) = L 1 H 2L1 - L 2 H 2L2 oil L 1 (2.3) 22

39 2.7.2 H2 consumption data reported in literature The values reported in the literature can be used for a rapid estimation of H2 consumption; values for each hydrotreating reaction are shown in Table 2.5 and these values differ between each other Stoichiometric H2 consumption in hydrotreating reactions Hydrogen consumption can be calculated based on the stoichiometry of each hydrotreating reaction. Lee et al. (2008) developed a method based on the stoichiometric H2 consumption during hydrotreating of a mixture of 62 vol. % straight-run heavy gas oil, 10 vol. % coker light gas oil and 28 vol. % FCC light cycle oil for the production of diesel taking into account HDS, HDN, HDO, HGO (olefins saturation in terms of bromine number) and HDA reactions. Hydrogen consumption obtained by this method was N m 3 H2/m 3 oil. The method presented a standard deviation of 20.7 N m 3 H2/m 3 oil in comparison with the gas streams analysis. Meanwhile, the analysis of the liquid streams presented a lower deviation (8.7 N m 3 H2/m 3 oil). Lee et al. (2008) method is shown below. H 2 consumption (N m3 H 2 m 3 ) = sg oil f [S f - S p sg p sg f Y p ] sg f [N f - N p sg p sg f Y p ] sg f [O f - O p sg p sg f Y p ] [Br f - Br p sg p sg f Y p ] (2.4) sg f [PNA f - PNA p sg p Y sg p + 3 [MA f - MA p sg p Y f sg p ]] f where: S, N, O, Br, PNA, and MA are the sulfur, nitrogen, oxygen, bromine number, polynuclear aromatics, and mononuclear aromatics contents in wppm, respectively; Sg is the specific gravity; Y is the total liquid product yield. The subscripts f and p represent the feed and products, respectively. 23

40 Stratiev et al. (2009) reported a similar method for hydrogen consumption of HDS, HDN HGO, and HDA reactions during hydrotreating of different feedstocks. The method is shown in equations Castañeda et al. (2011) compared experimental H2 consumption with different methods reported in the literature; this work noticed that hydrogen consumption calculated by Lee et al.(2008) and Stratiev et al. (2009) methods worked relatively well with middle fractions obtained during hydrotreating. In contrast, the methods presented a variation of up to 29.2 % when applied to heavy gas oil. H 2 consumption (N m3 H 2 m 3 ) = H oil 2 HDS + H 2 HDN + H 2 HDA + H 2 HGO (2.5) 1000 H 2 HDS = [3 (S f 32 - S Y p BT p p 320 ) + 2 (S 1000 DBT f f 32 Y p DBT p - S p )] ρ f 100 (2.6) H 2 HDN = [5 ( N f Y p N f 140 )] 22.4 ρ f 100 (2.7) H 2 HGO = [(O f 1000 Mw f - O p 10Y p Mw p )] 22.4 ρ f 100 (2.8) H 2 HDA = [[2 (PNA f 1000 Mw f - PNA p 10 Y p Mw p ) + 3 (MA f 1000 Mw f - MA p 10Y p Mw p )] 22.4 ρ f 100 ] Y p Y p + [[2 (TriA f - TriA Mw p ) + 2 (TriA f Mw f - TriA p p 100 ) + 2 (DiA 1000 f - Mw f (2.9) DiA p 10Y p Mw p ) + 3 (MA f - MA p 10Y p Mw p )] 22.4 ρ f 100 ] where: H2HDS, H2HDN, H2HGO, H2HDA, are the amount of H2 consumed in HDS, HDN, HGO, and HDA reactions, respectively. S, BT, DBT,N,O,PNA, MA,TriA, and DiA represent sulfur, benzothiophene, dibenzothiophene, nitrogen, olefins, polycyclic arenes, monoaromatics, tricyclic 24

41 arenes, dyciclic arenes, and monocyclic arenes contents, respectively (wt. %).ρ and Y is the density at 20 C (gr/ml) and the liquid yield (volumetric fraction), respectively. f and p represent feed and products, respectively. Hisamitsu et al. (1976) developed a method that only requires the characterization of the liquid streams and it is based on the decrease in aromatic content. This method also considers HDS and HDN reactions resulting in the prediction of theoretical H2 consumption. This method is adequate when limited hydrocracking is present. This method is shown below (Hisamitsu et al., 1976; Mapiour et al., 2010). H 2 consumption ( m3 H 2 m 3 ) = [(C A) f - (C A ) p ] ρ feed 22.4 oil ( [(S) f - (S) p ] ρ feed ) ( [(N) f - (N) p ] ρ feed ) (2.10) where: CA, S, and N are the aromatic carbon, sulfur and nitrogen contents (wt. %) respectively; 22.4 is the number of standard liters is a mole of an ideal gas. The subscripts f and p represent the feed and products, respectively. It can be seen in the equation that H2 consumption for sulfur and nitrogen conversions is multiplied by two. This represents the hydrogen required to form hydrocarbons during HDS and HDN, and the hydrogen content in the form of H2S and NH3 in the product Kinetic modelling This method is developed by analyzing the rates of all the reactions that consume hydrogen during hydrotreating. Therefore, it can provide accurate te values of H2 consumption. Papayannakos & Georgiou 1988 proposed a simple kinetic model without considering diffusion in the catalysts particles for HDS of an atmospheric residue in a trickle bed reactor at the following conditions: T = C; LHSV = h -1, P = 5 MPa. Co-Mo/Al2O3 commercial catalysts were used. They obtained the following equation from a differential mass balance based on an isothermal plug flow model: 25

42 (CH) 1-α (CHT) 1-α = (α-1) ξ (1-ε) kv (VR/QL) (2.11) Where: CH = remaining H2 demand or concentration of the bonds which are bound to react with H2 at reaction conditions (mol H2/m 3 oil); α = reaction rate order; CHT = concentration of all the bonds likely to react with H2 before any treatment occurs (mol H2/m 3 oil); ξ = Catalyst remaining relative activity; ε = Catalyst bed void fraction; kv = Rate constant; VR = Catalyst bed volume; and QL = Oil volumetric flow rate. The authors obtained a total H2 reaction rate order of two. Also, CH is the subtraction of CHT from the H2 consumption measured experimentally (CON). Therefore, eq takes the form of the following: 1 ξ CON (C HT -CON) C HT = (1-ε) k v V R Q L (2.12) The H2 consumption obtained by the model was mol H2/m 3 oil compared to mol H2/m 3 oil, which is based on the API of the residue. Moreover, they also confirmed that the value obtained by the model can approach experimental values by comparing it with H2 consumption of a 99.2 % HDS of a vacuum residue with similar properties as that of the feedstock. The H2 consumption obtained in the latter case was mol H2/m 3 oil Analysis of hydrogen consumption studies Methods based on stoichiometry and kinetics shown above calculate chemical H2 consumption during hydrotreating. The overall hydrogen consumption in the industry involves other factors such as mechanical losses due to the use of compressors, venting losses in case of purging the gas stream, and solution losses which are defined as the amount of hydrogen dissolved in the final product (Hisamitsu et al., 1976). Therefore, in the case of experimental or pilot plant studies, total H2 consumption must consider chemical H2 consumption and dissolved hydrogen. Castañeda et al. (2011) reported that the estimation of total H2 consumption could have an error of about 6% if dissolved H2 is not considered. 26

43 As was mentioned previously, H2 consumption is a dependent variable. One of the main factors affecting this variable is feedstock properties. Industries that work with hydrotreating units are concerned in the amount of hydrogen needed in the process due to the high costs and importance of this reactant. When refineries process one specific type of feedstock, the overall H2 balance will not show significant changes; however, when different feedstocks with different properties need to be treated, the industry may overestimate the usage of reactant causing economic losses. Furthermore, under usage of hydrogen may cause operational problems and possible shutdown of the unit. The models found in literature can be used for quick estimations, however, when more precision is needed these are not of great use e.g., the models were inaccurate to predict H2 consumption for feeds that were not used in the development of such models as it was mentioned previously. Also, the use of some models may be difficult e.g. models based on the stoichiometric H2 consumption on each type of reaction require detailed characterization of feedstocks and products.the development of regression models based on process conditions for feedstocks with varying composition and properties may contribute with important data to perform corrections in H2 global mass balances of refineries. 2.8 Vapor-liquid equilibrium in hydrotreating process The study of vapor-liquid interactions can contribute to the improvement of process design and catalyst formulations (Chávez et al., 2014; Chen et al., 2011). Reaction rates are dependent on the amount of hydrogen dissolved into the liquid which certainly will be affected by several factors such as H2 consumption, levels of conversion, feedstock properties, temperature, pressure, and liquid and gas flowrates. The general parameters of concern are H2 partial pressure, feed vaporization, and dissolved H2 (Mapiour et al., 2010). Vapor-liquid equilibrium is explained by a series of laws and principles that are briefly reviewed in this section. The principle of equilibrium in vapor-liquid interaction is similar to reversible chemical reactions; Gibbs energy (G) always tends to remain constant (G=0). Suppose there is an ideal mixture of decane and hydrogen in a flash operation, increasing temperature will increase the molar composition of decane in the gas phase modifying the partial pressure of each component 27

44 and forcing more hydrogen to dissolve in the liquid phase to force the system back into equilibrium (Mapiour et al., 2010; Skogestad, 2009).Three laws for ideal mixtures are the basis for this example. These laws are displayed in Table 2.6. Hydrocarbons and hydrotreating products (H2S, NH3) are not ideal under hydroprocessing conditions; therefore, calculation of vapor-liquid equilibria requires the use of equations of state developed for non-ideal mixtures. Peng-Robinson is widely used to predict the behavior of hydrocarbons mixtures-hydrogen at high temperature and pressures (Lal et al., 1999). The modification of this model developed by Stryjek and Vera (PRSV) can improve the prediction of the equilibrium when there is the presence of polar compounds in the system (Ghosh, 1999) Inlet and outlet hydrogen partial pressure Hydrogen partial pressure improves hydrotreating conversions by increasing the concentration of hydrogen in the liquid which will diffuse into the catalyst prior to reaction (Fogler, 1999). Optimal levels of H2 partial pressure will contribute to hydrotreating economically and will prevent the formation of coke; excessive levels of H2pp will have no significant impact on hydrotreating activities (Speight, 1999). Many studies have analyzed the effect of H2pp on hydrotreating conversions. The general trend is that H2pp improves hydrogenation reactions (Girgis & Gates, 1991); this means that HDA and HDN conversions will increase with increasing H2pp. In the case of HDS, H2pp also improves the conversion; however, the effect is less significant than in the other two reactions (Fang, 1999). In the hydrotreating process, outlet H2pp is lower than inlet H2pp due to the following factors: the presence of other gases in the system (H2S, NH3, C1-C4), H2 consumption, feed vaporization and dissolved hydrogen. The study of outlet H2pp during hydrotreating may be more relevant than inlet H2pp because it represents the average conditions in the catalyst bed and the last opportunity of the catalyst to react with the feedstock (Mapiour, 2009). However, studies on the effect of outlet H2pp on hydrotreating reactions and processing conditions are not readily available in the literature. 28

45 Table 2.6 Laws explaining liquid-gas interaction for ideal mixtures Author Description Equation Dalton Henry Raoult The partial pressure of a specific component is proportional to the mole fraction of the component multiplied by the system pressure. The amount of a gas dissolved in a liquid is proportional to the partial pressure of the gas in the system. The partial pressure of a specific component is proportional to the vapor pressure of the component multiplied by its mole fraction in the liquid phase. where: P=total pressure P = P 1 +P 2 +P 3.Pn Pn = X n P Pn= partial pressure of component n nx=mole fraction of component n H = n N λρ L where: H = Henry s coefficient nn = molar gas volume at standard conditions λ = solubility coefficient; and ρl = density of the liquid at process conditions y i P = x i P i sat (T) where: yi = mole fraction of component i in the gas phase P = total pressure x I = mole fraction of component i in the liquid phase P i sat (T) = vapor pressure of component i McCulloch & Roeder (1976), studied the effect of processing conditions on H2pp (inlet and outlet) at the following process conditions: T = C, P = psi, and G/O= scf/bbl. The feedstock properties were the following: density = 0.87g/cm 3, average boiling point = 296 C, and molecular weight of 224 g/mole. The authors provided a method to calculate inlet and outlet H2pp and studied the effect of process conditions on outlet H2pp; they determined that temperature increased feed vaporization and H2/oil ratio had a positive effect on outlet H2pp only. Mapiour et al. (2010), studied how process conditions interact with feed vaporization, 29

46 dissolved H2, H2 consumption and H2pp during hydrotreating of bitumen-derived heavy gas oil (HGO) at the following conditions: T = C, LHSV = h -1, and P = 7-11 MPa. The authors found out that temperature had a positive effect on feed vaporization and a slight negative effect on dissolved H2. Increasing H2pp leads to higher H2 dissolution and H2 consumption, and higher flow rates increased the amount of vaporized feed and dissolved H2. This work did not find any correlation between H2pp and feed vaporization. Therefore, experiments with lighter feeds could give an insight of how feed vaporization interacts with H2pp Calculation of hydrogen partial pressure Hydrogen partial pressure can be calculated by performing vapor-liquid equilibrium simulations on HYSYS software. The procedure to calculate H2pp is mentioned in Mapiour (2009) and it was adapted from McCulloch & Roeder (1976). The procedure is summarized below. First, the feedstock properties were entered in HYSYS. The feed can be simulated using the following properties: boiling point distribution, density, and sulfur and nitrogen content; the feedstock will approach more to reality when more properties are included. Several thermodynamic models have been used to develop vapor-liquid equilibria calculations for the interaction of hydrogen and hydrocarbons. Peng-Robinson (PR) thermodynamic model is widely used for these high pressure and high temperature systems (Lal et al., 1999); however, since there is the presence of polar compounds (H2S and NH3) on the gas outlet stream, PRSV model may improve the prediction of vapor-liquid equilibria (Ghosh, 1999). Second, use a flash unit in the simulation environment. Input temperature, pressure, and gas and liquid flow rates of each component. Inlet hydrogen partial pressure calculation requires hydrogen and feedstock streams data only. The outlet hydrogen partial pressure calculation takes into consideration the consumption of H2 and the properties and flow rates of the gases formed during the process (H2S, NH3, and light hydrocarbons). Once the simulation is performed, feed vaporization and hydrogen solubility can be calculated with the data provided by the simulation Dissolved hydrogen Dissolved hydrogen is proportional to the hydrotreating reactions rates; therefore, the study of this variable is important for an improvement in the design of hydrotreating units. Different 30

47 works related to reactor simulation and modeling have found out that when VLE are considered, the models improve the prediction of hydrotreating conversions (Akgerman & Netherland, 1986; Chen et al., 2011). Most of the studies related to dissolved hydrogen have been performed with simple hydrocarbons; however, data of real hydrotreating feedstocks at high pressures and temperatures is not readily available in the literature (Chávez et al., 2014; Lal et al., 1999). The solubility of hydrogen is affected by temperature, pressure, and boiling point distribution (BPD) of petroleum fractions (Riazi & Roomi, 2007). Many authors reported the following: as temperature and pressure increase, the mole fraction of hydrogen in the liquid also increases. In the case of petroleum fractions with higher boiling point distribution, the solubility of H2 in the latter tends to resemble dissolved H2 in lighter fractions at high temperatures (Cai et al., 2001; Florusse et al., 2003; Lal et al., 1999). 31

48 3 EXPERIMENTAL This chapter describes the experimental setup and procedure. Also, it describes the steps that were taken to carry the different phases of the study and the instruments used to characterize feed and products. 3.1 Materials Commercial Ni-Mo/γ-alumina catalyst (C424) employed in this work was provided by Criterion Catalysts. Silicon carbide and glass beads were supplied by Fisher Scientific, Edmonton, Canada. Butanethiol was purchased from Sigma-Aldrich Canada. Helium and hydrogen gases were bought from Praxair, Saskatoon, Canada. Light gas oils were supplied by Syncrude Research Centre, Edmonton, Canada. The properties of the light gas oils are displayed in Table 3.1. Table 3.1 Properties of feedstocks Property KLGO HLGO VLGO PHTHGO Density (g/cm 3 ) Sulfur content (wt. %) Nitrogen content (wt. %) Aromatics content (wt. %) Hydrogen-to-carbon (H/C) ratio Boiling point distribution ( C) * Hydrogen-to-carbon ratio for VLGO was taken from Yui,

49 3.2 Experimental setup The experimental setup is shown in Figure 3.1 and consists on a 304 stainless steel microreactor (inner diameter of 14mm and length of 240mm), a hydrogen mass flow controller to regulate the amount of H2 going into to the system, an electrical furnace with temperature control, a water scrubber to remove ammonium sulfide, a high pressure separator, a back pressure regulator, a bubble flow meter connected to the gas outlet, and a sodium hydroxide scrubber connected to the gas outlet to prevent release of hydrogen sulfide to the environment. PG FC Bubble flow meter (gas flowrate) PG PG PG CHECK VALVE REACTOR F U R N A C E PG TC PG BACK PRESSURE REGULATOR GC-FID (C1-C4 contents) TO VENT H2S SCRUBBER H2 He FEED TANK WATER SCRUBBER HIGH PRESSURE SEPARATOR NEEDLE VALVE PRODUCT STORAGE TANK Stripping NS analyzer(nitrogen and sulfur content) NMR(aromatic content) CHNS analyzer( hydrogen content) Figure 3.1 Experimental setup (PG: pressure gauge; FC: mass flow controller; TC: temperature controller) 3.3 Experimental procedure Before catalyst loading, the NiMo/γAl2O3 catalyst has to be dried at 200 C for three hours. Then, the reactor is loaded considering its division into three sections: bottom, middle and top. The top section in the upper part of the catalyst bed has one layer of 3 mm of glass beads and several layers of silicon carbide: 25, 10 and 10 mm height of 16, 46 and 60 mesh, respectively. The catalyst bed in the middle part consists of 90 mesh silicon carbide and 5 ml of catalyst mixed uniformly. The catalyst bed will have a height of around 10 cm. Finally, the bottom section is arranged similarly to the top section in reverse direction (Mapiour, 2009; Owusu-Boakye, 2005). 33

50 After loading the reactor, the pressure in the system was maintained at 8.96 MPa for 24 h to ensure a leakage free system. Subsequently, the reactor temperature was increased to 100 C and sulfidation of the catalyst began by pumping 100 ml of butanethiol solution (2.9 vol. % in heating oil) at a flow rate of around 2.5 ml/min to wet the catalyst. Then, H2 was introduced at H2/oil ratio of 600 m 3 /m 3 and flow rate was reduced to achieve LHSV = 1h -1. The temperature of the catalyst bed was then gradually increased to 193 C and maintained for 24 h. Then, the temperature was again increased gradually to 343 C and maintained for another 24 h. Following sulfidation, the catalyst was precoked by treating LGO at 370 C for a period of 5 days; pressure, LHSV, and H2/oil ratio were maintained constant. Samples were collected every 12 hours, stripped with nitrogen, and then analyzed for nitrogen, sulfur and aromatics conversions (Botchwey et al., 2003). 3.4 Experimental plan Phase I - Development of H2 consumption regression models during hydrotreating of gas oils Experimental runs were designed based on a central composite design (CCD) methodology in MINITAB 7.0 software. CCD is used to build quadratic models with the objective to optimize process variables and study the interaction between and within variables and specific responses. CCD consists of one central point (repeated six times to improve accuracy), two axial points (2n), and two cube points (2 n ) in which n are the number of the experimental variables (Lazic, 2004). In case of having three input variables, the number of experimental runs to be carried are twenty as shown in equation 3.1. N= 2 n +2n+6 = 20 (3.1) The experimental variables and their ranges for this study are the following: T = C, P = MPa, and LHSV = h -1 ; H2 /oil ratio was maintained constant at 600 Nm 3 H2/m 3 oil. The experimental design consists of twenty runs in which six are repeated to improve the model accuracy. The experimental matrix for VLGO, HLGO, KLGO, and PHTHGO feedstocks employed in this work is displayed in Table

51 Hydrogen consumption was measured by analyzing the H2 content in gas and liquid streams. Then, these data were compared with an approach developed by Hisamitsu et al. (1976). The best method (between gas and liquid) was selected to perform statistical analysis and build H2 consumption regression models for each feedstock based on process conditions. Then, a composite model was developed by the statistical analysis of a new set of data. This set of data was obtained for calculating total H2 consumption of this mixture by assuming equal volumetric flow rate of each of four streams: VLGO, HLGO, KLGO, and PHTHGO in the gas oil mixture. The prediction of the composite model was tested with a mixture of LGOs within the range of conditions provided by the CCD. The conditions were the following: T = 365 and 375 C, P = 9.31 MPa, LHSV = 1.2 h -1, and H2/oil ratio = 600 m 3 /m 3. The response of the model was also compared with values reported by Edgar (1993), and models developed by Hisamitsu et al. (1976) and Lee et al. (2008). These models available in the literature predict chemical H2 consumption only; therefore, dissolved H2 was calculated and added to obtain the total H2 consumption. Dissolved H2 was calculated by VLE simulations in HYSYS In addition, data of H2 inlet pp, H2 outlet pp, and feed vaporization obtained from the simulation was employed to study the effects of process conditions on these variables and the correlation between these variables. It is important to note that for the control experiment, the calculation of these four parameters was done by taking an average of hydrogen consumption Phase II - Effects of process conditions on HDS, HDN, and HDA conversions during hydrotreating of gas oils and optimization of process conditions HDS, HDN, and HDA conversions were calculated for each experimental run in Table 3.2. For each experimental run, two samples were taken every 12 hours for reproducibility, and when conditions changed there was a stabilization period of 24 hours. Also, to ensure that catalyst deactivation did not occur, hydrotreating conversions were compared between the first and last three experiments carried at the central conditions. Then, statistical analysis was carried out to study interactions between and within hydrotreating conversions and process conditions, and develop regression models. The regression models were employed to find the best sets of conditions to process each feedstock. 35

52 Table 3.2 Experimental matrix Run T P LHSV HDS, HDN, HDA (%); H2 consumption (Nm 3 H2/m 3 oil) (⁰C) (MPa) (h- 1 ) VLGO HLGO KLGO PHTHGO

53 3.5 Characterization of feedstocks and products Boiling point distribution Simulated distillation to calculate boiling point distribution was carried out in a Varian CP 3800 gas chromatographer (GC) coupled to a Varian CP8400 autosampler for injection. The basic principle of this analysis is to separate the components of the mixture based on their boiling points. The volatilized sample is carried by He (mobile phase) at 30mL/min through a capillary column in which the components of the sample are separated based on their partition time. The column has the following dimensions: length: 10 m; diameter: 0.53 mm; nominal film thickness: 0.88 mm. The temperature of the oven and detector were 380 C and 375 C, respectively. Following separation, the components are analyzed by a flame ionization detector (FID) which combust the gasses in the presence of H2 (35 ml/min) and air (400 ml/min) releasing specific ions that are proportional to the concentration of the species in the gas mixture (Owusu-Boakye, 2005) Nitrogen and sulfur analysis Nitrogen and sulfur analysis were carried in an Antek 9000model: 9000NS. The nitrogen contents were analyzed by a combustion/chemiluminescence technique following ASTM 4629 procedure. The nitrogen groups are oxidized to nitric oxide (NO) at temperatures higher than 1000 C. Then, these groups react with ozone to produce metastable nitrogen dioxide species (NO2*). When the latter species begin to decay, they emit energy in the form of light (chemiluminescence) which is detected by a photomultiplier at a specific wavelength (Owusu- Boakye, 2005). The sulfur contents were analyzed by a combustion/fluorescence technique following ASTM 5463 procedure. When the sulfur contents of the sample are oxidized they are converted into sulfur dioxide (SO2). The latter is then excited by UV light to produce an unstable SO2 *. When SO2 * starts to decay into its stable form, there is an emission of fluorescence which is detected by a photomultiplier (Mapiour, 2009). The chemistry of nitrogen and sulfur analysis is shown below. 37

54 R-N + R-S + O 2 CO 2 + H 2 O + NO + SO 2 (3.2) SO 2 + hv ' SO 2 + hv '' (3.3) NO + O 3 NO 2 * NO 2 + hv (3.4) The sulfur and nitrogen contents are calculated based on the following equation: C x (ppm) = (I - Y) S M * K g (3.5) where: Cx is the concentration of sulfur or nitrogen species; I is the average integrated detector response for test specimen solution (counts); Y is the y-intercept of standard curve (counts); S is the slope of standard curve (counts/mg); M * is the mass of test specimen solution injected; and Kg is the gravimetric dilution factor (g/g) Analysis of aromatics content Total aromatics content was analyzed by 13 C nuclear magnetic resonance spectroscopy at 500MHz in Fourier transform mode. The conditions for the analysis were inversed gate decoupling, sweep width of 27.7 khz and a pulse delay of 4 seconds. Analysis of each sample took around 1.5 h for 2000 scans (Owusu-Boakye, 2005). 13 C has two orbital spins (½ and -½) in the presence of an external magnetic field. The latter has to be strong to obtain a valuable energy difference between these two spins. The samples for this analysis are placed in a special glass and radio frequencies (RF) are broadcasted through a coil that is situated around the glass. Each chemical species will absorb the RF to a different extent; therefore, the variation between the incoming and out coming frequencies allow the analysis of the composition of the sample. Tetramethylsilane is used as a standard to obtain a chemical shift to differentiate between saturated carbon (0-50ppm) and aromatic carbon ( ppm) (Owusu- Boakye, 2005). The aromatic contents can be calculated by the following equation: 38

55 C Ar = I Ar I Ar + I Sat 100 (3.6) where: CAr is the aromatics content; IAr is the the integral of total aromatics; Isat is the integral of total saturates Measurements of hydrogen content Hydrogen contents in liquid products and feedstocks were analyzed with an elemental analyzer. Samples are combusted at 1000 C in excess of oxygen to form carbon dioxide and water; these products are then taken out from the combustion chamber by helium and passed through a copper metal to remove the oxygen that was not consumed during combustion. Finally, the compounds are quantified by a TCDs (thermal conductivity detectors) Analysis of light hydrocarbons The gas stream produced during hydrotreating was collected in a tedlar bag and then analyzed in a 7890A gas chromatograph. The instrument contains a complex valve system to carry the sample through five packed columns and one capillary column. In addition, the equipment contains a flame ionization detector (FID) for light hydrocarbons and a thermal conductivity detector (TCD) for hydrogen. Gas chromatography was used to measure the amount of feed cracked during hydrotreating in the form of C1-C4. 39

56 4 RESULTS AND DISCUSSION 4.1 Development of hydrogen consumption regression models during hydrotreating of gas oils Hydrogen global mass balance The global hydrogen mass balance (see equation 2.1) was calculated based on the hydrogen content in both gas and liquid streams. The global mass balances were calculated for KLGO and HLGO in the experiments carried at LHSV = 1 h -1 and LHSV = 2h -1 shown in the experimental matrix (Table 3.2). The mass balance for these experiments was in the range of %. The data and results of the hydrogen mass balances are displayed in the appendices Experimental hydrogen consumption H2 consumption was determined by the analysis of H2 content in liquid and gas streams at laboratory conditions; the laboratory average conditions were the following: T =23 C and P = 96.0 kpa. The experimental data of H2 consumption by gas analysis were transformed to normal conditions (T = 20 C and P = kpa) using equation 4.1 and will be reported as N m 3 H2 / m 3 oil. V N = V P P N T N T (4.1) where: V, P, and T are H2 consumption, pressure, and temperature at laboratory conditions in m 3 H2 / m 3 oil, kpa, and K, respectively. The subscript N represents normal laboratory conditions. The major source of experimental error for the determination of H2 consumption by gas and liquid analyses was related to to the gas flow rates and hydrogen concentrations, respectively. 40

57 H 2 consumption (Hisamitsu approach) (N m 3 H 2 /m 3 oil) Gas flow rates measured by a bubble flow meter presented an error of ± 3 ml/min which translated into a standard deviation of ± 2.1 N m 3 H2/m 3 oil for H2 consumption by gas analysis; flow rate measurement was carried out six times for each experimental run. In the case of H2 consumption by liquid analysis, H2 concentrations of liquids determined by the elemental analyzer varied significantly and the standard deviation of H2 consumption by liquid analysis was ± 27 N m 3 H2/ m 3 oil; H2 concentrations were measured 2 times for each experimental run. The liquid analysis could have been inaccurate for the following reasons: H2 tendency to escape from the products during stripping, the experimental setup could have retained some liquid in the section above the separator, and the elemental analyzer had issues to carry the samples to the combustion section. H2 consumption results based on gas and liquid analyses were compared with a method reported in the literature which determines H2 consumption based on changes in the aromatics content; it is well known that most of the hydrogen is consumed by HDA. For instance, Owusu- Boakye (2005), removed 10.8 wt. % of aromatics in comparison with 1.68% of sulfur from an LGO blend at the following conditions: T ( C), P ( MPa), LHSV ( h- 1 ), and H2/oil= 500 m 3 / m 3. The method used for the comparison was shown in section and it was adapted from Hisamitsu et al. (1976). The comparisons were performed with the experiments and feedstocks used for the H2 global mass balances; the comparisons are shown in Figure 4.1 for HLGO and Figure 4.2 for KLGO Gas-Hisamitsu Liquid-Hisamitsu Linear (Gas-Hisamitsu) Linear (Liquid-Hisamitsu) R² = R² = H 2 consumption (gas and liquid) (N m 3 H 2 /m 3 oil) Figure 4.1 Comparison of methods to calculate H2 consumption for HLGO 41

58 H 2 consumption (Hisamitsu approach) (N m 3 H 2 /m 3 oil) Gas-Hisamitsu Liquid-Hisamitsu Linear (Gas-Hisamitsu) Linear (Liquid-Hisamitsu) R² = R² = H 2 consumption (gas and liquid) (Nm 3 H 2 /m 3 oil) Figure 4.2 Comparison of methods to calculate H2 consumption for KLGO The data points shown in Figures 4.1 and 4.2 describe the relationships for H2 consumption determined in the present study with gas analysis-hisamitsu approach and liquid analysis- Hisamitsu approach. It can be seen in Figures 4.1 and 4.2 that the coordinates of the axes do not match for either of the correlations which means that these methods provide different predictions of H2 consumption. However, there is a better agreement between gas analysis from the present study with Hisamitsu approach. Based on the comparison between H2 consumption methods and the standard deviations associated with gas and liquid analysis, the data and statements from this point forward regarding H2 consumption are based on the gas analysis. Normal hydrogen consumption for the experiments given by the CCD are shown in Table 4.1 and there is a standard deviation of ±1.4 N m 3 H2/m 3 oil between the control experiments. 42

59 Run Table 4.1 H2 consumption by gas analysis during hydrotreating of VLGO, HLGO, T (⁰C) P (MPa) LHSV (h- 1 ) KLGO, and PHTHGO H2 consumption (N m 3 H2 / m 3 oil) VLGO HLGO KLGO PHTHGO

60 H 2 consumption (N m 3 H 2 /m 3 oil) Hydrogen consumption is affected by feedstock properties as shown in Figure 4.3. KLGO had maximum H2 consumption among the feedstocks due to highest sulfur content and along with PHTHGO, these feeds had a higher content of aromatics than the other feeds. In case of VLGO and HLGO, it can be seen that VLGO consumed more H2 than HLGO. Even though HLGO has higher aromatics content than VLGO, the latter may result in having less aromatics content in its products. Moreover, VLGO feedstock contains around 2 times more sulfur content than HLGO. PHTHGO feed consumed less H2 in comparison with the others due to the pre-treatment of the feed which could have saturated most of the polyaromatics content. Thus, majorly leaving monoaromatics in the PHTHGO feed; it is well known that monoaromatics are more stable than polyaromatics and are hardly saturated at normal hydrotreating conditions.furthermore, PHTHGO has less sulfur content that the other three feedstocks. In general, H2 consumption had the following order: KLGO>VLGO>HLGO>PHTHGO. 160 KLGO HLGO VLGO PHTHGO Experimental runs provided by CCD Figure 4.3 H2 consumption comparison among feedstocks 44

61 4.1.3 Hydrogen consumption regression models Statistical analysis was performed to develop regression models for H2 consumption based on hydrotreating process variables. The experimental data of H2 consumption was submitted to a test of significant factors. This test defines that a single factor or interaction between and within factors is significant when the p-value is lesser than 0.05 at a 95% confidence level. Then, the models were evaluated by an R 2 test which describes the variability of the model with respect to the experimental data. R 2 values range from 0 to 1; as R 2 approaches 1, the experimental data representation improves; this value increases along with the number of predictors. On the other hand, adjusted R 2 (Adj R 2 ) only improves if the variables or interactions are significant in the response of the model. Therefore, adjusted R 2 (Adj R 2 ) is most commonly used (Lazic, 2004). It is important to mention that the models presented in this work were obtained from different trials in which the factors or interactions were either eliminated or included to achieve the best value of Adj R 2. Furthermore, the models also considered T, P, or LHSV single terms if there was any interaction within or between these variables.the p-value test for H2 consumption is shown in Table 4.2 and the empirical models for each feed are shown in Table 4.3. Table 4.2 p-value test; H2 consumption during hydrotreating of VLGO, HLGO, KLGO and PHTHGO Factor or interaction p-value of factor or interaction VLGO HLGO KLGO PHTHGO T <0.01 <0.01 < P <0.01 < LHSV 0.05 <0.01 < (T) 2 - <0.01 < (P) (LHSV) T * P T * LHSV P*LHSV

62 Table 4.3 H2 consumption regression models for VLGO, HLGO, KLGO and PHTHGO Feed Model R 2 Adj R 2 VLGO H2 consumption (N m 3 H2/m 3 oil) = T P LHSV (P 2 ) HLGO H2 consumption (N m 3 H2/m 3 oil) = T P LHSV (T 2 ) (P LHSV) KLGO H2 consumption (N m 3 H2/m 3 oil) = T LHSV (T 2 ) PHTHGO H2 consumption (N m 3 H2/m 3 oil) = T P LHSV (T 2 ) (P 2 ) (LHSV 2 ) The experimental data of H2 consumption for all the feeds was used to build a general model; it was assumed that in case of a mixture, each feed contributes to the total H2 consumption in the same proportion. The statistical analysis of the combined data results in the model shown in equation 4.2. This composite model has an Adj R 2 of 0.98 and a standard error of ± 1.02 N m 3 H2/m 3 oil. H 2 consumption (N m3 H 2 m 3 ) = T P LHSV oil (4.2) (T 2 ) (P 2 ) (LHSV 2 ) (T LHSV) (P LHSV) The composite model (equation 4.2) was tested experimentally and compared with models and values reported in the literature. The feedstock for the experiments was a mixture of VLGO, KLGO, and HLGO at the same volumetric proportion. Two experiments at conditions not provided by the CCD were performed for this purpose. The conditions were the following: T = 365 C and 375 C, P = 9.31 MPa, LHSV = 1.2h -1, and H2/oi l= 600 N m 3 /m 3. The mixture had a density of 0.88 gr/cm 3, 1.56 % sulfur, % nitrogen, and % aromatics. The hydrotreating conversions and dissolved H2 during hydrotreating of the mixture are shown in Table

63 Table 4.4 HDS, HDN, and HDA conversions, and H2 consumption during hydrotreating of LGOs mixture. Exp. Experimental conditions HDS HDN HDA Dissolved H2 Run T( C) P(MPa) LHSV(h -1 ) (%) (%) (%) (N m 3 H2/m 3 oil) The experimental H2 consumption of the mixture was compared with the three models and values reported in the literature. Hisamitsu et al. (1976) and Lee et al. (2008) models were shown in section 2.7.3; meanwhile, Edgar(1993) published the following H2 consumption data for each hydrotreating reaction: HDS=13.4 N m 3 H2/m 3 oil per each 1 wt. % removed; HDN= 57.9 Nm 3 H2/m 3 oil per each 1 wt. % removed; HDA= 4.81 N m 3 H2/m 3 oil per each 1wt. % removed. The usage of the model developed by Lee et al. (2008) was made based on the following assumptions: HDO, olefins saturation and saturation of monoaromatics were not taken into account. Also, the density of the products was equal to the density of the feed, and liquid product yield was equal to 1. To improve the prediction of the models obtained from the literature, dissolved hydrogen during hydrotreating of VLGO, KLGO, and HLGO calculated on HYSYS was added to the chemical H2 consumption. The comparison is shown in Table 4.5 and it can be noted that the regression model was able to predict H2 consumption more precisely than the other three models. Table 4.5 Comparison between H2 consumption models for hydrotreating of petroleum fractions H2 consumption (Nm 3 H2/m 3 oil) Exp. Run Experimental H2 consumption (Nm 3 H2/m 3 oil) Composite regression model Edgar (1993) Hisamitsu et al. (1976) Lee et al. (2008)

64 4.2 Effects of temperature, pressure, and LSHV on hydrotreating conversions and optimization of process conditions during hydrotreating of gas oils In this section, the effects of processing variables on hydrotreating conversions were studied for VLGO, HLGO, KLGO and PHTHGO for the experimental runs provided by CCD. After carrying the experiments, regression models were developed to estimate the best sets of conditions to process each feedstock; giving maximum HDS, HDN, and HDA conversions Effects of temperature, pressure, and LSHV on HDA conversions Hydrodearomatization (HDA) reaction consumes more hydrogen than HDS and HDN during hydrotreating. Therefore, the effects of process conditions on this reaction were examined first. The experimental data of HDA conversions for each feed is shown in Table 4.6 and the standard deviation for HDA conversions was in the range of ± %; catalyst deactivation was not observed. The conversion of the data point in which maximum conversion was achieved for each feedstock into the amount of aromatics removed from each feed gives the following results: KLGO= 11.4%; VLGO=10.3%; HLGO=9.0%; PHTHGO= 6.1%. These results confirm the H2 consumption trend indicated in section The statistical analysis and regression models for HDA conversions are shown in Tables 4.7 and 4.8, respectively. It can be noted that the regression analysis shows no interaction between process variables; therefore, the analysis of the effects of process conditions on HDA was carried out by analyzing the experimental data at the following conditions: T = 353,370, and 387 C; P = 7.80, 8.96, and MPa; LHSV = 0.7, 1.5, and 2.3 h -1 as shown in Figures

65 Table 4.6 HDA conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO Run T (⁰C) P (MPa) LHSV (h- 1 ) HDA (wt. %) VLGO HLGO KLGO PHTHGO

66 Table 4.7 p-value test for HDA conversions during hydrotreating of VLGO, HLGO, KLGO and PHTHGO Factor or interaction p-value of factor or interaction VLGO HLGO KLGO PHTHGO T <0.01 <0.01 < P <0.01 < LHSV - <0.01 < (T) 2 - <0.01 <0.01 <0.01 (P) (LHSV) T * P T * LHSV P*LHSV Table 4.8 HDA regression models for VLGO, HLGO, KLGO, and PHTHGO Feed Model R 2 Adj KLGO HDA (%) = T LHSV (T 2 ) R 2 HLGO HDA (%) = T P LHSV (T 2 ) VLGO HDA (%) = T P PHTHGO HDA (%) = T P + 99 LHSV (T 2 ) (P 2 ) (LHSV 2 ) (T P) (T LHSV) 0.94 (P LHSV)

67 HDA (%) HDA (%) 50.0 VLGO HLGO KLGO PHTHGO Temperature C Figure 4.4 Effects of temperature on HDA conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Pressure and LHSV constant at 8.96 MPa and 1.5 h -1, 50.0 VLGO HLGO KLGO PHTHGO Pressure (MPa) Figure 4.5 Effects of pressure on HDA conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Temperature and LHSV constant at 370 C and 1.5 h -1, respectively 51

68 HDA (%) VLGO HLGO KLGO PHTHGO LHSV (h -1 ) Figure 4.6 Effects of LHSV on HDA conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Temperature and pressure constant at 370 C and 8.96 MPa, respectively. HDA is a reaction limited at high temperatures due to the reversibility nature of hydrogenation (Gary et al., 2007). The equilibrium will shift in such a way that the reaction will produce more aromatic compounds instead of producing saturation (Owusu-Boakye et al., 2006). Temperature limits can be observed for each feedstock in Figure 4.4. Owusu-Boakye et al. (2006) found out that the optimum temperature to carry HDA during hydrotreating of an LGO blend was 379 C. Meanwhile, Chandra-Mouli & Dalai (2009) noted that this negative effect of temperature on HDA conversions began at 300 C when processing pre-hydrotreated LGO with an aromatics content of 16.4 % wt. Specific temperature limits will be reported in the optimization section of this work. In case of LHSV, it can be observed in Figure 4.6 that increasing LHSV led to a decrease in HDA conversions for all the feedstocks. Mapiour et al. (2010) found a similar trend during hydrotreating of heavy gas oil under similar conditions. In contrast, when statistical analysis was performed, LHSV did not show a significant effect on HDA conversions for VLGO and PHTHGO. Mann et al. (1987) also found out that residence time had a non-significant effect on HDA 52

69 conversions during hydrotreating of coker naphtha at the following conditions: T = C, P = MPa, and LHSV= h -1. The role of pressure on HDA conversions for this study is discussed below. It can be seen in Figure 4.5 that increasing pressure did not show an improvement on HDA of KLGO at MPa and in case of PHTHGO, increasing pressure did not improve aromatics saturation at any of the conditions employed in this work. These effects may be due to the aromatic groups present in these feeds. KLGO originates from coking processes in which big amounts of aromatics compounds and olefins are produced (Aoyagi et al., 2003). Furthermore, coking processes vacuum residue which may contain high contents of monoaromatics. These compounds may not be completely saturated at the conditions employed in this work. In case of PHTHGO, the primary hydrotreating process could have saturated most of the polyaromatics compounds which result in a lower HDA efficiency for this study. On the other hand, pressure had a positive effect on HDA conversions during hydrotreating of HLGO and VLGO; HDA is favored by high H2pp due to the hydrogenation mechanism taken by aromatic compounds Effects of temperature, pressure, and LSHV on HDS of gas oils The experimental data for HDS for each feedstock is shown in Table 4.9 and the standard deviation for HDS conversions was in the range of ± %. The effects of process conditions on HDS were studied in two forms. First, the independent or single effects of temperature, pressure, and LSHV on HDS conversions were studied by analyzing the experimental data at the following conditions: T = 353,370, and 387 C; P = 7.80, 8.96, and MPa; LHSV = 0.7, 1.5, and 2.3 h -1 ; the graphical representation of independent effects of process conditions on HDS conversions is presented as final sulfur content in products (sulfur removal) in terms of ppm for a better visuality of the trends; the standard deviation in terms of ppm was ± ppm. Second, in case of interaction between variables provided by the regression models, Minitab software was used to develop graphs that show the corresponding interactions. The statistical analysis and regression models for HDS conversions are shown in Tables 4.10 and 4.11, respectively. 53

70 Run Table 4.9 HDS conversions during hydrotreating of VLGO, HLGO, KLGO, and T (⁰C) P (MPa) LHSV (h- 1 ) PHTHGO HDS (wt. %) VLGO HLGO KLGO PHTHGO

71 Table 4.10 p-value test for HDS activities during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO Factor or interaction p-value of factor or interaction VLGO HLGO KLGO PHTHGO T 0.02 <0.01 <0.01 <0.01 P LHSV <0.01 < <0.01 (T) (P) (LHSV) T * P T * LHSV < P*LHSV Table 4.11 HDS regression models for VLGO, HLGO, KLGO, and PHTHGO Feed Model R 2 Adj R 2 KLGO HDS (%) = T P LHSV (T 2 ) (T P) (P LHSV) HLGO HDS (%) = T P LHSV VLGO HDS (%) = T LHSV (T LHSV) PHTHGO HDS (%) = T LHSV

72 S content in products (ppm) The processing variables affect gas oils to different extents depending on their composition and properties. An increase in temperature improved sulfur removal for all the feedstocks as it can be seen in Figure 4.7. Similarly, the temperature was a significant factor in the statistical analysis for all the feedstocks as shown in Table This general trend has been observed by many works related to hydrotreating of several petroleum fractions (Yui & Chan, 1992; Bej et al., 2007). Yui & Chan (1992) reported that during hydrotreating of coker naphtha with a NiMo catalyst at T = C, P = 3-5 MPa, LHSV = 1-2 h -1 and G/O=600 m 3 /m 3, HDS conversions increased with an increase in temperature. In addition, the authors found that pressure did not have any significant effect on HDS conversions VLGO HLGO KLGO PHTHGO Temperature C Figure 4.7 Effects of temperature on sulfur removal during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Pressure and LHSV constant at 8.96 MPa and 1.5 h -1, respectively Figure 4.8 shows that the positive effect of pressure on sulfur removal was only clear for HLGO. The latter statement is also supported by the p-value test. In case of KLGO, the statistical analysis shows a significant interaction between pressure and temperature on HDS conversions. 56

73 S content in products (ppm) This effect is shown graphically in Figure 4.9. The surface plot in Figure 4.9 (a) shows that changes in both temperature and pressure at constant LHSV had an impact on HDS conversions. These plots may be difficult to read; therefore, it is advised to include an interaction plot which is shown in Figure 4.9 (b); interaction terms are significant when the trend lines of an interaction plot are not parallel. Girgis & Gates (1991) stated that HDS reaction may proceed by two mechanisms: direct scission of the C-S bond by hydrogenolisis, or hydrogenation of an unsaturated bond followed by hydrogenolisis, the former reaction is reversible. This means that the removal of sulfur for KLGO could have been limited by hydrogenation due to the temperature limits that are shown in Figure 4.9. Moreover, NiMo catalyst is reported to achieve high hydrogenation activities (Aoyagi et al., 2003; Knudsen et al., 1999) VLGO HLGO KLGO PHTHGO Pressure (MPa) Figure 4.8 Effects of pressure on sulfur removal during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Temperature and LHSV constant at 370 C and 1.5 h -1, respectively 57

74 S urface Plot of HDS(%) vs P(MPa), T(⁰C) 100 HDS(%) T(⁰C) 9 P(MPa) (a) (b) Figure 4.9 Surface plot (a) and Interaction plot (b) of effects of temperature and pressure on HDS conversions at LHSV = 1.5 h -1 for KLGO 58

75 S content in productss (ppm) In case of LHSV, Figure 4.10 shows that any increase in LHSV had a negative effect on the sulfur removal for all the feeds. This effect is due to less residence time of the reactants in the reactor (Bej et al., 2007; Botchwey et al., 2003). Moreover, the regression analysis displayed in Table 4.10 indicates that temperature and LHSV had a joint effect on HDS conversions for VLGO. According to Figure 4.11, HDS is taking place in excess at LHSV= 1 h -1 ; therefore, any further increase in temperature has a negative effect on HDS conversions. Meanwhile, at higher LHSV, the temperature continues to have a positive effect on HDS because the reaction is not being limited by equilibrium. The regression analysis also indicates that LHSV did not have any effect on HDS conversions for KLGO; however, Figure 4.10 shows the contrary. It is important to note that the trends provided by the models are not completely accurate; the models have an Adj R 2 in the following range: VLGO HLGO KLGO PHTHGO LHSV (h -1 ) Figure 4.10 Effects of LHSV on sulfur removal during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Temperature and pressure constant at 370 C and 8.96 MPa, 59

76 Figure 4.11 Interaction plot of the effects of temperature and LHSV on HDS conversions during hydrotreating of VLGO. Pressure constant at 8.96 MPa Effects of temperature, pressure, and LSHV on HDN conversions The experimental data for HDN conversions during hydrotreating of each feedstock is shown in Table 4.12 and the standard deviation for HDN conversions was in the range of ± %. The statistical analysis and regression models for HDN conversions are shown in Tables 4.13 and Similar to HDS conversions, the independent effects of process conditions on HDN were studied with the experimental data at the same conditions as mentioned in section Meanwhile, the interactions between process variables were studied by analyzing the trends provided by the HDN regression models. 60

77 Table 4.12 HDN conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO Run T (⁰C) P (MPa) LHSV (h- 1 ) HDN (wt. %) VLGO HLGO KLGO PHTHGO

78 Table 4.13 p-value test for HDN conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO Factor or interaction p-value of factor or interaction VLGO HLGO KLGO PHTHGO T <0.01 <0.01 <0.01 <0.01 P <0.01 <0.01 <0.01 <0.01 LHSV <0.01 <0.01 < (T) < (P) (LHSV) < T * P < T * LHSV < P*LHSV < < Table 4.14 HDN regression models for VLGO, HLGO, KLGO, and PHTHGO Feed Model R 2 Adj R 2 KLGO HLGO VLGO PHTHGO HDN (%) = T P LHSV (T 2 ) (P 2 ) (LHSV 2 ) (T P) (T LHSV) (P LHSV) HDN (%) = T P LHSV (T P) (T LHSV) HDN (%) = T P LHSV (T P) (T LHSV) (P LHSV) HDN (%) = T P LHSV (LHSV 2 ) (T P) (T LHSV)

79 HDN (%) In general, HDN conversions improve with increasing temperature and pressure. Meanwhile, any decrease on the residence time will affect HDN conversions negatively (Girgis & Gates, 1991). As it can be observed in Figures , HDN conversions for all the feedstocks followed this general trend. In addition, the interaction between processing variables affected HDN conversions for VLGO, KGLO, and PHTHGO. Figures 4.15 and 4.16 show that changes in temperature and pressure were more significant at higher values of LHSV to compensate the lesser residence time. Pressure and temperature effects for KLGO shown in Figure 4.15 plateau at high levels of conversion. This means that the reaction passes through a maximum due to equilibrium limitations; HDN takes place by hydrogenation followed by C-N bond scission (Angelici, 1997) VLGO HLGO KLGO PHTHGO Temperature ( C) Figure 4.12 Effects of temperature on HDN conversions during hydrotreating of VLGO, HLGO, KLGO, and PHTHGO. Pressure and LHSV constant at 8.96 MPa and 1.5 h -1, respectively 63